About

I’m interested in the gap between computation and cognition — why today’s AI systems can compute remarkably well, yet still fall short of the persistent, adaptive, and embodied intelligence we associate with human cognition.

My interests span representation learning, probabilistic modeling, active inference, world models, and hierarchical learning systems. I care about how intelligent systems construct internal models of themselves and the world, reason under uncertainty, and continually adapt from experience.

Previously, I worked on applied machine learning problems involving weak supervision, noisy labels, ambiguous targets, and scalable model development. I’m currently building a stealth startup exploring what computing looks like when intelligence becomes continuous rather than session-based.

I left Meta to buy myself time to study intelligence full-time, funding that decision by building an artisanal furniture business. The ideas that emerged during those years became the foundation for the company I’m building today.